• The site broke after I enabled a rule. Safe Mode — from the admin bar, wherever you are — then revert the rule on VelocityForge > Rollback. If the admin will not load, add define('VELOCITYFORGE_PANIC', true); to wp-config.php.
• I want to know the escape hatches work before I need them. Run the recovery drill on VelocityForge > Rollback. It presses the real ones, including a loopback on the genuine secret URL, and restores every value and the on-disk secret file exactly as it found them.
• I deleted the plugin and the site is still broken. It should not be: uninstall restores autoload backups first. If the white screen predates VelocityForge, check the error log — the panic constant only disables our own mutations.
• A rule disappeared and I got an e-mail about it. That is the Pro rule watch: the day after the rule started applying, LCP or INP was worse than your threshold, so it was reverted and the reason auto-watch was written on the rollback event. Pin the rule if you want it kept regardless, or raise the threshold, or lengthen the window.
• The watch says “inconclusive”. It saw fewer measurements than the sample floor (30 by default), or Safe Mode was on, or the beacon is off. The rule stays exactly as it was; nothing is reverted on thin evidence.
• An update slowed the site and I do not know why. Open the regression row’s “What changed” — it names the scripts and stylesheets the front end gained, lost or grew, read from a snapshot of the real page taken before the files were replaced.
• A card says “theme” instead of a plugin name. The owner resolver only claims what it can support: a prefix it knows, a handle this site enqueues, or the velocityforge/interaction_owner filter. Where none of those answer it says “theme” or nothing rather than guessing.
• An element is missing from the INP card. It needs at least three measurements in the last 28 days. Below that a 75th percentile is not a percentile.
• Critical CSS looks wrong on one page. It is matched by URL pattern; add that URL as its own record, or disable the feature — the original stylesheet always remains as the fallback.
• No field data. The beacon reports from real visits, not from admin page views. A quiet site takes a few days.
• Numbers differ from PageSpeed Insights. They should: this is field data from your visitors, PSI is a lab run on one simulated device. With Pro’s scheduled lab audits switched on you get both, as two lines on one chart, and they are never compared against each other.
• The TBT budget never triggers anything. It cannot on field data — Total Blocking Time needs a long-tasks trace the beacon does not capture. It is read by the Pro lab audits’ comparison only.
• A test alert says the address is refused. The channel URL must be https and must not resolve to a private or loopback address; the send-test runs the same SSRF guard as the save form, on purpose.
• My alert secret shows as ••••1234. That is the whole point — secrets are encrypted at rest and never echoed back. Leave the field empty to keep the stored value.
• The Speculation panel is on but view-source shows no rules. WordPress never speculates for a logged-in user — check in a private window — and it switches speculative loading off entirely while permalinks are set to Plain. The panel names whichever of the two applies.
• VelocityForge stopped optimising after I installed another cache plugin. By design: one of the seventeen recognised plugins is active and our overlapping work stood down. The notice lists every feature that stepped back, not only the obvious one.